EP0855110A1 - An adaptive echo cancellation method - Google Patents

An adaptive echo cancellation method

Info

Publication number
EP0855110A1
EP0855110A1 EP96935668A EP96935668A EP0855110A1 EP 0855110 A1 EP0855110 A1 EP 0855110A1 EP 96935668 A EP96935668 A EP 96935668A EP 96935668 A EP96935668 A EP 96935668A EP 0855110 A1 EP0855110 A1 EP 0855110A1
Authority
EP
European Patent Office
Prior art keywords
echo
filter
erl
constant
time average
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
EP96935668A
Other languages
German (de)
French (fr)
Other versions
EP0855110B1 (en
Inventor
Johnny Karlsen
Anders Eriksson
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Telefonaktiebolaget LM Ericsson AB
Original Assignee
Telefonaktiebolaget LM Ericsson AB
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Telefonaktiebolaget LM Ericsson AB filed Critical Telefonaktiebolaget LM Ericsson AB
Publication of EP0855110A1 publication Critical patent/EP0855110A1/en
Application granted granted Critical
Publication of EP0855110B1 publication Critical patent/EP0855110B1/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B3/00—Line transmission systems
    • H04B3/02—Details
    • H04B3/20—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other
    • H04B3/23—Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other using a replica of transmitted signal in the time domain, e.g. echo cancellers

Definitions

  • the present invention relates to an adaptive echo cancellation method in which adaption of an echo canceller is prevented in an environment with low signal to background noise ratio.
  • Echo is a problem related to the perceived speech quality in telephony systems with long delays, e.g. telephony over long distances or telephony systems using long processing delays, like digital cellular systems.
  • the echo arises in the four-to-two wire conversion in the PSTN/subscriber interface.
  • echo cancellers are usually provided in transit exchanges for long distance traffic, and in mobile services switching centers for cellular applications.
  • the echo canceller Due to the location of the echo canceller it i ⁇ made adaptive,- the same echo canceller i ⁇ used for many different subscriber ⁇ in the PSTN. This adaption is neces ⁇ ary not only between different calls, but also during each call, due to the non-fixed nature of the tran ⁇ mis ⁇ ion network, e.g. pha ⁇ e ⁇ lip ⁇ , three-party calls, etc .
  • the threshold is made adaptive in [1] .
  • the method of [1] is based on comparison of the power of the input signal and the background noise level .
  • the adaption is prevented if the power of the echo (power of input ⁇ ignal - echo path attenuation ERL) is less than the background noise level plus a margin of 1 to 5 dB.
  • a problem with the described approach is its dependence on an accurate estimate of the echo path attenuation ERL. If a large value of ERL is estimated, the adaption may be completely inhibited. Hence, the filter coefficients are frozen, and, assuming ERL is estimated from the filter coefficients, no new estimate of ERL will be found. If the characteristics of the echo generating system now change, the filter will not be able to adapt to the new situation. Thus, the method suggested in [1] is too conservative, i.e. filter updating is inhibited also in situations where this should be avoided.
  • An object of the present invention i ⁇ to provide an adaptive echo cancellation method in which adaption of an echo canceller i ⁇ prevented in an environment with low ⁇ ignal to background noise ratio only when absolutely neces ⁇ ary.
  • FIGURE 1 is a block diagram of an echo generating sy ⁇ tem,-
  • FIGURE 2 i ⁇ a block diagram of an echo cancellation system
  • FIGURE 3 i ⁇ a time diagram of the input ⁇ ignal power to an echo canceller; and FIGURE 4 is a flow chart of a preferred embodiment of the method in accordance with the present invention.
  • Fig. 1 illustrates the echo generating process in a telephony system.
  • a subscriber A called the far end subscriber below, is connected to a hybrid (a hybrid forms the interface between a four-wire and a two-wire connection, as is well known in the art) over a two-wire line.
  • a subscriber B called the near end subscriber below, is connected to another hybrid over a two- wire line.
  • the two-wire lines transfer both incoming and outgoing speech signals.
  • Outgoing speech from far end subscriber A i ⁇ transferred to near end subscriber B over the upper two-wire line in Fig. 1.
  • outgoing speech from near end sub ⁇ criber B is transferred to far end subscriber A on the lower two-wire line in Fig. 1.
  • the lower two-wire line from ⁇ ub ⁇ criber B to subscriber A also contains an echo of outgoing speech from subscriber A, which the hybrid at sub ⁇ criber B wa ⁇ not able to ⁇ uppre ⁇ s completely.
  • the upper two-wire line in Fig. 1 contains echo from outgoing speech from subscriber B.
  • Fig. 2 illustrate ⁇ how the echo back to subscriber A is cancelled at the near end side (a similar arrangement i ⁇ provided at the far end side) .
  • Input signal x(n) represents speech from subscriber A.
  • input ⁇ ignal x(n) is also forwarded to an adaptive filter, which models the impulse response of the hybrid by adjusting its filter coefficients (a typical filter length i ⁇ 512 coefficient ⁇ ) .
  • the re ⁇ ulting e ⁇ timate of echo signal ⁇ (n) i ⁇ denoted s(n) .
  • This estimate is subtracted from output signal y(n) , and the resulting error signal e(n) is forwarded to the adaptive filter for adjustment of the filter coefficients and to the two-wire line back to far end subscriber A.
  • the echo s(n) is modelled using an FIR (Finite Impulse Response) model, and the estimate s(n) is determined by the normalized least mean square (NLMS) method (see e.g. [2]) .
  • NLMS normalized least mean square
  • Ev 2 (n) is the variance of the near end noise v(n) (signal from near end subscriber B during time periods without speech) .
  • the error in estimate s (n) is due to errors in the estimated FIR coefficients ⁇ b k ⁇ .
  • Fig. 3 illustrates the above described situation in time diagram form.
  • Curve 1 in Fig. 3 represents the input power R x of input signal x(n) .
  • the signal is rather strong and well above the noise level NL (v(n) with no near end speech), represented by the dash-dotted line in Fig. 3.
  • the filter would converge even in the valleys of the curve.
  • the distance between noise level NL and the signal is reduced, either due to a lower signal power as represented by curve 2 in Fig. 3 or by a higher noise level NL, the filter will diverge in the valleys.
  • the method described in [1] compares the power of the input signal x(n) to the background noise level NL.
  • An adaption of the filter is prevented if the power of the echo s (n) is less than the background noise level, with a margin of 1 to 5 dB. That is,
  • ERL denotes an estimate of the echo path attenuation and C is a constant safety margin (in the range of 1 to 5 dB) .
  • a problem with the approach described above is its dependence on an accurate estimate of the echo path attenuation ERL. If a large value of ERL is estimated, the adaption may be completely inhibited, since R X /ERL may very well be below C-NL for all input signal levels. Hence, the filter coefficients are frozen, and, as ⁇ uming ERL i ⁇ estimated as the sum of the squares of the filter coefficients, no new estimate of ERL will be formed. Hence, condition (3) is too conservative.
  • the filter may end up in an adaption dead-lock. This may for in ⁇ tance be the case for an input signal similar to curve 3 in Fig. 3, which lies completely under the dashed line ERL+NL+C (ERL-NL-C expressed in dB) .
  • the basic idea in accordance with the present invention to overcome this problem is to control updating of the filter by comparing a short time average R x sta of the power of x(n) to a long time average R x lta of the power of x(n) . If the short time average falls below the long time average, filter adaption is inhibited.
  • D is a predefined constant (for example of the order of -45 dBmO) used to inhibit adaption during periods of both low input power and low background noise levels, and where ⁇ is given by
  • the constant or (>0) ensures that adaption of the filter is never completely inhibited (when the estimated value of ERL is large, ⁇ will be chosen in equation (5) , since in this case ⁇ will be less than ERL- L/R x lta ) .
  • o/ ⁇ l for example 0.95
  • the filter is at least updated when the short time average R x ⁇ ta of the input signal x(n) exceeds the long time average R x lta .
  • the method in accordance with the present invention is illustra- ted in curve 3.
  • the short double arrow R x sta represent ⁇ the short time average.
  • the length of the double arrow represent ⁇ the time interval over which the average i ⁇ formed (typical value ⁇ are 60- 70 milli ⁇ econds) .
  • the double arrow designated R x lca represents the long time average, which typically is computed over a time interval that is at lea ⁇ t an order of magnitude longer than the time period for calculating the short time average, for example of the order of 4 seconds. Thu ⁇ , only the mo ⁇ t recent ⁇ ample ⁇ are u ⁇ ed for calculating the short time average, while a large number of ⁇ ample ⁇ are u ⁇ ed for calculating the long time average.
  • the short time average at sample instant n exceeds the long time average at the same instant (the di ⁇ tance above the t-axis represent ⁇ the corre ⁇ ponding average) .
  • Thu ⁇ in this case (curve 3, sample instant n) the present invention would allow filter updating, while the method in accordance with the prior art would inhibit filter updating.
  • step 10 the next sample of input ⁇ ignal x(n) i ⁇ collected.
  • step 12 a new long time average R lca including the new sample is calculated.
  • step 14 a new short time average R x sta including the new sample is cal ⁇ culated.
  • step 16 ⁇ is calculated in accordance with (5) above.
  • step 16 a reference level R is calculated in accordance with the right hand side of relation (4) above.
  • step 20 the short time average is compared to this reference. If the short time average falls below the reference level, filter updating is inhibited in step 24, otherwise the filter is updated in step 22. Thereafter the algorithm returns to step 10 for collecting the next sample.
  • D may be set to 0, which corresponds to - ⁇ dBmO or zero background noise.
  • the right hand side of (4) will always equal ⁇ R x lta .

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
  • Filters That Use Time-Delay Elements (AREA)
  • Telephone Function (AREA)

Abstract

An adaptive echo cancellation method determines a long time aver age (12) of an input signal and a short time average (14) of the same signal, and prevents updating of an adaptive filter if the short time average (14) falls below (24) the long time average.

Description

AN ADAPTIVE ECHO CANCELLATION METHOD
TECHNICAL FIELD
The present invention relates to an adaptive echo cancellation method in which adaption of an echo canceller is prevented in an environment with low signal to background noise ratio.
BACKGROUND OF THE INVENTION
Echo is a problem related to the perceived speech quality in telephony systems with long delays, e.g. telephony over long distances or telephony systems using long processing delays, like digital cellular systems. The echo arises in the four-to-two wire conversion in the PSTN/subscriber interface. To remove this echo, echo cancellers are usually provided in transit exchanges for long distance traffic, and in mobile services switching centers for cellular applications.
Due to the location of the echo canceller it iε made adaptive,- the same echo canceller iε used for many different subscriberε in the PSTN. This adaption is necesεary not only between different calls, but also during each call, due to the non-fixed nature of the tranεmisεion network, e.g. phaεe εlipε, three-party calls, etc .
A problem with this filter adaption procesε iε that the filter may diverge if the input εignal decreaseε to a level that approaches the background noise level. To prevent the filter from diverging in situations where the background noise level is comparable to the signal level, it haε been suggeεted [1] to inhibit updating of the filter when the power of the input signal iε leεs than a given threshold. To overcome problems related to uεing a fixed threshold, the threshold is made adaptive in [1] . The method of [1] is based on comparison of the power of the input signal and the background noise level . The adaption is prevented if the power of the echo (power of input εignal - echo path attenuation ERL) is less than the background noise level plus a margin of 1 to 5 dB.
A problem with the described approach is its dependence on an accurate estimate of the echo path attenuation ERL. If a large value of ERL is estimated, the adaption may be completely inhibited. Hence, the filter coefficients are frozen, and, assuming ERL is estimated from the filter coefficients, no new estimate of ERL will be found. If the characteristics of the echo generating system now change, the filter will not be able to adapt to the new situation. Thus, the method suggested in [1] is too conservative, i.e. filter updating is inhibited also in situations where this should be avoided.
SUMMARY OF THE INVENTION
An object of the present invention iε to provide an adaptive echo cancellation method in which adaption of an echo canceller iε prevented in an environment with low εignal to background noise ratio only when absolutely necesεary.
This object is solved by the featureε of claim 1.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention, together with further objects and advantageε thereof, may best be understood by making reference to the following description taken together with the accompanying drawingε, in which:
FIGURE 1 is a block diagram of an echo generating syεtem,-
FIGURE 2 iε a block diagram of an echo cancellation system;
FIGURE 3 iε a time diagram of the input εignal power to an echo canceller; and FIGURE 4 is a flow chart of a preferred embodiment of the method in accordance with the present invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Fig. 1 illustrates the echo generating process in a telephony system. A subscriber A, called the far end subscriber below, is connected to a hybrid (a hybrid forms the interface between a four-wire and a two-wire connection, as is well known in the art) over a two-wire line. Similarly a subscriber B, called the near end subscriber below, is connected to another hybrid over a two- wire line. The two-wire lines transfer both incoming and outgoing speech signals. Outgoing speech from far end subscriber A iε transferred to near end subscriber B over the upper two-wire line in Fig. 1. Similarly outgoing speech from near end subεcriber B is transferred to far end subscriber A on the lower two-wire line in Fig. 1. However, the lower two-wire line from εubεcriber B to subscriber A also contains an echo of outgoing speech from subscriber A, which the hybrid at subεcriber B waε not able to εuppreεs completely. Similarly the upper two-wire line in Fig. 1 contains echo from outgoing speech from subscriber B.
Fig. 2 illustrateε how the echo back to subscriber A is cancelled at the near end side (a similar arrangement iε provided at the far end side) . Input signal x(n) , where n denotes diεcrete time, represents speech from subscriber A. The input εignal x(n) is attenuated by the hybrid (the attenuation is represented by the echo path attenuation ERL (ERL = Echo Return Loss) ) , and the resulting echo signal s(n) is combined with the near end εignal v(n) , which may or may not contain near end speech. Thuε, the resulting output signal y(n) contains both the near end signal and echo from the far end signal. Furthermore, input εignal x(n) is also forwarded to an adaptive filter, which models the impulse response of the hybrid by adjusting its filter coefficients (a typical filter length iε 512 coefficientε) . The reεulting eεtimate of echo signal ε (n) iε denoted s(n) . This estimate is subtracted from output signal y(n) , and the resulting error signal e(n) is forwarded to the adaptive filter for adjustment of the filter coefficients and to the two-wire line back to far end subscriber A.
Often the echo s(n) is modelled using an FIR (Finite Impulse Response) model, and the estimate s(n) is determined by the normalized least mean square (NLMS) method (see e.g. [2]) . For time invariant signals it can be shown that the steady-state misadjustment, i.e. the power of the error of the estimated echo, E(s(n) -s(n) )2, for a constant step-size μ of the NLMS method equals (see e.g. [2])
E( § (n) - s (n) ) 2 = —H_.sv2(n) (1)
2-μ
where Ev2(n) is the variance of the near end noise v(n) (signal from near end subscriber B during time periods without speech) . However, the error in estimate s (n) is due to errors in the estimated FIR coefficients {bk} . These FIR filter coefficient errors may be approximated by (based on eq. (45) in [2] )
B S> -b skJ, * . - rX . l . BvXnL (2)
* 2-μ N Ex2 (n) where Ν is the filter length. Assuming that the filter haε converged in a stationary scenario, the variance of the filter coefficientε is given by (2) . Now, if the power of input signal x(n) decreases, (2) yields an increased steady-state variance of the filter coefficients, and the filter will diverge. If the power of x(n) increases again, the filter will re-converge, but the estimation error E(ε (n) -ε (n) )2 may be undesirably high before the filter has re-converged. Hence, some sort of control of the update process of the filter is desirable in order to prevent the estimation error from increasing too drastically in situations with non-stationary input signal characteristics.
Fig. 3 illustrates the above described situation in time diagram form. Curve 1 in Fig. 3 represents the input power Rx of input signal x(n) . In this case the signal is rather strong and well above the noise level NL (v(n) with no near end speech), represented by the dash-dotted line in Fig. 3. In this case the filter would converge even in the valleys of the curve. However, if the distance between noise level NL and the signal is reduced, either due to a lower signal power as represented by curve 2 in Fig. 3 or by a higher noise level NL, the filter will diverge in the valleys. As mentioned above, to overcome this problem the method described in [1] compares the power of the input signal x(n) to the background noise level NL. An adaption of the filter is prevented if the power of the echo s (n) is less than the background noise level, with a margin of 1 to 5 dB. That is,
— < C'NL (3)
ERL where ERL denotes an estimate of the echo path attenuation and C is a constant safety margin (in the range of 1 to 5 dB) .
A problem with the approach described above is its dependence on an accurate estimate of the echo path attenuation ERL. If a large value of ERL is estimated, the adaption may be completely inhibited, since RX/ERL may very well be below C-NL for all input signal levels. Hence, the filter coefficients are frozen, and, asεuming ERL iε estimated as the sum of the squares of the filter coefficients, no new estimate of ERL will be formed. Hence, condition (3) is too conservative. The filter may end up in an adaption dead-lock. This may for inεtance be the case for an input signal similar to curve 3 in Fig. 3, which lies completely under the dashed line ERL+NL+C (ERL-NL-C expressed in dB) .
The basic idea in accordance with the present invention to overcome this problem is to control updating of the filter by comparing a short time average Rx sta of the power of x(n) to a long time average Rx lta of the power of x(n) . If the short time average falls below the long time average, filter adaption is inhibited.
This basic idea may be formalized as follows: the adaption is inhibited if
where D is a predefined constant (for example of the order of -45 dBmO) used to inhibit adaption during periods of both low input power and low background noise levels, and where γ is given by
The constant or (>0) ensures that adaption of the filter is never completely inhibited (when the estimated value of ERL is large, α will be chosen in equation (5) , since in this case < will be less than ERL- L/Rx lta) . By choosing o/~l (for example 0.95) , the filter is at least updated when the short time average Rx βta of the input signal x(n) exceeds the long time average Rx lta.
The method in accordance with the present invention is illustra- ted in curve 3. The short double arrow Rx sta representε the short time average. The length of the double arrow representε the time interval over which the average iε formed (typical valueε are 60- 70 milliεeconds) . Similarly the double arrow designated Rx lca represents the long time average, which typically is computed over a time interval that is at leaεt an order of magnitude longer than the time period for calculating the short time average, for example of the order of 4 seconds. Thuε, only the moεt recent εampleε are uεed for calculating the short time average, while a large number of εampleε are uεed for calculating the long time average. As can be seen from the figure the short time average at sample instant n exceeds the long time average at the same instant (the diεtance above the t-axis representε the correεponding average) . Thuε, in this case (curve 3, sample instant n) the present invention would allow filter updating, while the method in accordance with the prior art would inhibit filter updating.
A preferred embodiment of the method in accordance with the present invention will now be described with reference to the flow chart in Fig. 4. In step 10 the next sample of input εignal x(n) iε collected. In step 12 a new long time average R lca including the new sample is calculated. Similarly, in step 14 a new short time average Rx sta including the new sample is cal¬ culated. In step 16 γ is calculated in accordance with (5) above. In step 16 a reference level R is calculated in accordance with the right hand side of relation (4) above. In step 20 the short time average is compared to this reference. If the short time average falls below the reference level, filter updating is inhibited in step 24, otherwise the filter is updated in step 22. Thereafter the algorithm returns to step 10 for collecting the next sample.
In a simplified embodiment of the present invention D may be set to 0, which corresponds to -∞ dBmO or zero background noise. In this case the right hand side of (4) will always equal γRx lta.
It will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departure from the spirit and scope thereof, which is defined by the appended claims.
REFERENCES
[1] WO93/09608, Nokia Telecommunications OY
[2] D.T.M. Slock, "On the convergence behavior of the LMS and the normalized LMS algorithms", IEEE Transactions on Signal Processing, 41 (9) :2811-2825, September 1993.

Claims

1. An adaptive echo cancellation method in which adaption of an echo canceller is prevented in an environment with low signal to background noise ratio, including the steps of determining an echo path attenuation (ERL) estimate and a noise level estimate (NL) of said environment, characterized by the steps of: determining a long time average (Rx lta) of recent samples of an input signal x(n) to said echo canceller,- determining a short time average (Rx sta) of recent samples of said input signal x(n) ; preventing adaption of said echo canceller if said short term average iε less than the maximum of said long time average multiplied by a predetermined factor (γ) and a first predetermi¬ ned constant (D) .
2. The method of claim 1, characterized by said predetermined factor being the minimum of
(a) a second predetermined constant ( ) , and
(b) a product of said noiεe level estimate and said echo path attenuation eεtimate (ERL) divided by εaid long term average (Rx lta) .
3. The method of claim 2, characterized by εaid first predetermi¬ ned conεtant corresponding to a background noise level of the order of -45 dBmO .
4. The method of claim 2 or 3 , characterized by said second predetermined constant being approximately 1.
5. The method of claim 4, characterized by said second predeter¬ mined constant being equal to 0.95.
6. The method of any of the preceding claimε, characterized by said short term average being formed over a time period of the order of 60-70 ms .
7. The method of any of the preceding claims, characterized by said long term average being formed over a time period of the order of 4 seconds.
8. The method of claim 2, characterized by said first predetermi¬ ned constant being equal to 0, which corresponds to zero background noise.
EP96935668A 1995-10-11 1996-10-02 An adaptive echo cancellation method Expired - Lifetime EP0855110B1 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
SE9503545A SE505152C2 (en) 1995-10-11 1995-10-11 Adaptive echo extinguishing procedure
SE9503545 1995-10-11
PCT/SE1996/001239 WO1997014230A1 (en) 1995-10-11 1996-10-02 An adaptive echo cancellation method

Publications (2)

Publication Number Publication Date
EP0855110A1 true EP0855110A1 (en) 1998-07-29
EP0855110B1 EP0855110B1 (en) 2004-09-22

Family

ID=20399785

Family Applications (1)

Application Number Title Priority Date Filing Date
EP96935668A Expired - Lifetime EP0855110B1 (en) 1995-10-11 1996-10-02 An adaptive echo cancellation method

Country Status (10)

Country Link
US (1) US6137882A (en)
EP (1) EP0855110B1 (en)
JP (1) JP4027421B2 (en)
KR (1) KR100320315B1 (en)
CN (1) CN1098572C (en)
AU (1) AU708551B2 (en)
CA (1) CA2234555C (en)
DE (1) DE69633454T2 (en)
SE (1) SE505152C2 (en)
WO (1) WO1997014230A1 (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7423983B1 (en) * 1999-09-20 2008-09-09 Broadcom Corporation Voice and data exchange over a packet based network
US6351531B1 (en) * 2000-01-21 2002-02-26 Motorola, Inc. Method and system for controlling echo cancellation using zero echo path, ringing, and off-hook detection
JP4569618B2 (en) * 2006-11-10 2010-10-27 ソニー株式会社 Echo canceller and speech processing apparatus
US9344579B2 (en) 2014-07-02 2016-05-17 Microsoft Technology Licensing, Llc Variable step size echo cancellation with accounting for instantaneous interference
KR200481190Y1 (en) 2015-02-27 2016-08-26 최효묵 Game set of slap-match

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3789165A (en) * 1972-04-24 1974-01-29 Communications Satellite Corp Echo canceller with variable threshold
US4129753A (en) * 1977-12-09 1978-12-12 Bell Telephone Laboratories, Incorporated Echo canceller using feedback to improve speech detector performance
US4491701A (en) * 1981-03-05 1985-01-01 At&T Bell Laboratories Adaptive filter including a far end energy discriminator
US4405840A (en) * 1981-03-05 1983-09-20 Bell Telephone Laboratories, Incorporated Echo canceler far end energy discriminator
US4591669A (en) * 1984-09-26 1986-05-27 At&T Bell Laboratories Adaptive filter update gain normalization
US4712235A (en) * 1984-11-19 1987-12-08 International Business Machines Corporation Method and apparatus for improved control and time sharing of an echo canceller
US4922530A (en) * 1988-03-18 1990-05-01 Tellabs, Inc. Adaptive filter with coefficient averaging and method
US4918727A (en) * 1988-06-09 1990-04-17 Tellabs Incorporated Double talk detector for echo canceller and method
FI97657C (en) * 1991-11-04 1997-01-27 Nokia Telecommunications Oy A method for preventing adaptive echo cancellation divergence in a noisy signal environment
JP3094634B2 (en) * 1992-02-19 2000-10-03 日本電気株式会社 Echo removal method and echo removal device
DE4430189A1 (en) * 1994-08-25 1996-02-29 Sel Alcatel Ag Adaptive echo cancellation method
JP2947093B2 (en) * 1994-11-02 1999-09-13 日本電気株式会社 Method and apparatus for system identification with adaptive filters
US5631900A (en) * 1995-09-29 1997-05-20 Crystal Semiconductor Double-Talk detector for echo canceller

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See references of WO9714230A1 *

Also Published As

Publication number Publication date
CA2234555C (en) 2004-12-21
AU7380296A (en) 1997-04-30
AU708551B2 (en) 1999-08-05
EP0855110B1 (en) 2004-09-22
SE9503545D0 (en) 1995-10-11
JP4027421B2 (en) 2007-12-26
SE9503545L (en) 1997-04-12
KR100320315B1 (en) 2002-04-22
US6137882A (en) 2000-10-24
KR19990064190A (en) 1999-07-26
JPH11513553A (en) 1999-11-16
CN1098572C (en) 2003-01-08
WO1997014230A1 (en) 1997-04-17
CN1199519A (en) 1998-11-18
DE69633454D1 (en) 2004-10-28
DE69633454T2 (en) 2005-10-13
SE505152C2 (en) 1997-07-07
CA2234555A1 (en) 1997-04-17

Similar Documents

Publication Publication Date Title
EP0868787B1 (en) Method and device for echo cancellation using power estimation in a residual signal
AU710224B2 (en) An adaptive dual filter echo cancellation method
US4757527A (en) Echo canceller
US6516063B1 (en) Echo canceller having improved non-linear processor
US5631900A (en) Double-Talk detector for echo canceller
US5535194A (en) Method and apparatus for echo canceling with double-talk immunity
US4845746A (en) Echo canceller with relative feedback control
US20030076949A1 (en) Echo canceller employing dual-H architecture having improved non-linear echo path detection
US6137882A (en) Adaptive echo cancellation method
US5477535A (en) Method of preventing a divergence of an adaptive echo canceller in a noisy signal environment
US6266409B1 (en) Echo canceller employing dual-H architecture having improved double-talk detection

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): DE FI FR GB

17P Request for examination filed

Effective date: 19980318

17Q First examination report despatched

Effective date: 20030730

GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

GRAS Grant fee paid

Free format text: ORIGINAL CODE: EPIDOSNIGR3

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): DE FI FR GB

REG Reference to a national code

Ref country code: GB

Ref legal event code: FG4D

REF Corresponds to:

Ref document number: 69633454

Country of ref document: DE

Date of ref document: 20041028

Kind code of ref document: P

ET Fr: translation filed
PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed

Effective date: 20050623

REG Reference to a national code

Ref country code: FR

Ref legal event code: PLFP

Year of fee payment: 20

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: FI

Payment date: 20151028

Year of fee payment: 20

Ref country code: GB

Payment date: 20151027

Year of fee payment: 20

Ref country code: DE

Payment date: 20151028

Year of fee payment: 20

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: FR

Payment date: 20151019

Year of fee payment: 20

REG Reference to a national code

Ref country code: DE

Ref legal event code: R071

Ref document number: 69633454

Country of ref document: DE

REG Reference to a national code

Ref country code: GB

Ref legal event code: PE20

Expiry date: 20161001

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: GB

Free format text: LAPSE BECAUSE OF EXPIRATION OF PROTECTION

Effective date: 20161001